Professional Documents
Culture Documents
GEOLOGICAL SURVEY
by
This report was prepared under contract to the U.S. Geological Survey and has
not been reviewed for conformity with USGS editorial standards and strati-
Final Report
February 1981
The Electrical Properties of Clay
by
INTRODUCTION
rocks has not been investigated with the same intensity as have the effects of
ical community who have examined clay effects have been primarily concerned
(1957) and Madden and Marshall (1959) who describe, respectively, time-domain
Madden and Marshall (1959) was simplified and compared to measurements made on
current flow. This implies that electrical properties differ from place-to-
place within the rock, and that the clay distribution is not uniform when
carbon saturation from well logs of resistivity. Waxman and Smits (1972) and
Waxman and Thomas (1976) present a semi-empirical model for describing the
dependence of resistivity on clay content, expressed as cation exchange capac-
ity (CEC) per unit pore volume. They assume that resistivity is frequency
independent and that all of the clay in their samples contributed to the
nately there does not exist presently a sufficiently quantitative model for
clay-bearing rocks and mixtures of clay and glass beads. Our main objective
tion of zones with equal area but unequal transference number. It is assumed
that zone 1 contains pore fluid which is uninfluenced by clay and therefore
in near proximity to clay particles. The exchange cations associated with the
concentration of cations near the clay particle. Some of the anions, however,
are permanently bound to the clay mineral lattice. Thus, when current is
rather than anions. Madden and Marshall treat this enhanced conduction by
cations, thus lumping together the concentration and mobility for each type of
ion. Thus, their model loses any precise quantitative relationship with the
excess surface charge associated with the clay, which is related to the CEC
per unit pore volume of the Waxman-Smits model. Although the Madden-Marshall
model may present an adequate physical description, it is not amenable for use
ever, since it can provide some insight into the dependence of clay response
ary conditions between zones, Madden and Marshall derive a rather complicated
where
UjL = ionic mobility in zone 1,
F = Faraday constant,
D = diffusivity,
t = transference number,
co = angular frequency,
rlLl
xi
2Di
in which cations move unhindered and for which anions are completely blocked
resistive,
Z(o)
l B
A02 92
Vl
frequency for the indicated model parameters. The amplitudes have been nor-
previously. Figure 1 shows the effect of variation in Lj, the distance be-
quency, but otherwise remains unchanged. The amount of the shift is equal to
o
(Lj) . Figures 2 and 3 display the results of varying A and a ^. In each
case, the frequency effect shows a strong dependence, with only a slight
1000
Madden Marshal 1 vary LI
a
<
o:
.. 10
2-101
LOG FREQUENCY
a 2 - 0.01.
1000-
.. 100
.. 10
2-1012
LOG FREQUENCY
-3-2-1012
LOG FREQUENCY
Figure 3. Dependence of the Madden-Marshall model on ao Normalized ampli-
tive. This model then leads to the puzzling prediction that decreased clay
ness of the clay zones. This would sensibly predict that clay with increased
CEC would produce larger frequency effects. Because of the high negative
correlation between A and a> the Madden-Marshall model is, in general, inade-
The form of the plots in figures 1 through 3 suggests that Z(w) can be
R (1 - m (1 - - - )) (2)
(1 + JU>T>
where
a = 1/2,
R = background resistivity,
m = 1 - = chargeability,
K
O
f = time constant,
It is not possible to cast Z(o>) (equation (1)) exactly in the form of a Cole-
may constitute a more practical tool for characterizing clay response. From
the Madden-Marshall model, we could predict that the time constant T would
depend primarily on the distance between clay zones. Thus, T could provide
be expected to depend on both the amount of clay present in isolated zones and
the CEC. Clay which provided a continuous path for surface conduction, as
no effect on m.
ing section, we present data which were obtained with artifical clay-glass
bead mixtures and which have been inverted and interpreted in terms of a
10
1000
Invent Madden Manshal1 model
Madden-Marsna//
V
-3 -2 -1
LOG FREQUENCY
11
Measurements on Artificial Mixtures
with glass beads of varying size. In addition, the conductivity of the pore
solution was varied with the addition of NaCl. Measurements of amplitude and
phase were made mostly over the range of 1/128 to 1024 Hz using a ZONGE re-
ceiver. The usual procedure was to mix wet the desired quantity of clay with
glass beads. The mixture was then packed into sections of plexiglass tubing
after which fine nylon screen was fastened on the ends. The samples were
allowed to dry and were then vacuum-saturated with the desired solution. A
high humidity cell (fig. 5) was used to mount the samples such that 4-
(1 + (ju>T)C )a
For this model, the phase has slopes of c and -ac at low and high frequency,
respectively, on a plot of log ($) versus log (u)). The solid lines in figures
6 through 9 give the theoretical results for the best fit model in each
case. The asymmetric phase peaks and steep phase slopes at low frequency
12
\v )
Ag/AgCI kaotinite
electrodeX. contact * o--ring
1
N\ XI
; t i / ^^ 1
plaster plug / /^ !
current _
electrode foam ] piston
i
pad t
drain N support
13
1000
R = /0.6
m= 0.075
T-/.8
C = 0.72
0 = 0.38
20... . 100
I
O
.. 10
2-10123
LOG FREQUENCY
Figure 6. Observed amplitude and phase and inversion results for 3 percent
14
1000
fl - 13.6
m = o.oss
T r 4.45
C * 0.56
0 = 0.62
,20... ., 100
.. 10
6..
5.
-3-2-10123
LOG FREQUENCY
Figure 7. Observed amplitude and phase and inversion results for 3 percent
A/m2 .
15
1000
.. 100
t Q
X or
o
.. 10
-3 -2 -101
LOG FREQUENCY
Figure 8. Observed amplitude and phase and inversion results for 3 percent
16
1000
14.2
m* 0./0
7" 34.6
C s 0.60
0.36
5. 1
2-1 0 1 2,3
LOG FREQUENCY
Figure 9. Observed amplitude and phase and inversion results for 3 percent
9
Ca-raontmorillonite by weight, coarse glass beads, 0.01m NaCl, 0.1 A/m .
17
6r
IT
UJ
CD
j___I
.4 .6 .7 .8
C
8 r
(T
UJ
0 .2 .4 .6 .8 1.0
a
18
estimated values of the parameters a and c for all spectra measured on artifi-
cial samples for which the phase peak was well defined. As can be seen, all
of the estimated values of c are greater than 0.5 and most of the values of a
are less than 0.5. These results suggest that membrane polarization effects
can be physically described by the Madden-Marshall model and empirically
described by the generalized Cole-Davidson model. The fact that the observed
slopes are not exactly 1.0 and 0.5, as for the Cole-Davidson distribution
(fig. 4), could be due to a distribution of length scales associated with a
made in which the weight percent montmorillonite (measured dry) was fixed at
3.0, and the grain size of the glass beads was varied. Figures 6 through 9
display the observed spectra for 0.01m NaCl and the following grain sizes:
effect of increasing the grain size is to shift the relaxation to lower fre-
quency, thus increasing T. This result indicates that the length scale which
controls the time constant is dependent on the size of the glass beads rather
than the size of the clay particles, in agreement with results of Vacquier et
19
al (1957). This would tend to support the Madden-Mar shall model in which the
time constant is dependent on the distance between selective zones which, for
Figure 11 summarizes the results of varying the grain size for the samples
predicts that would depend on (L^) 2 . From figure 11, however, we tend to
the Madden-Marshall model predicts time constants which are close to observed
the montmorillonite, the plot of log RQ versus log a e would be a straight line
with slope of 1.0. The slopes in figure 12 of less than 1.0 for small a e are
evidence that the clay is contributing to the samples 1 conductivity. From
ing to aati 0.027 and 0.17 (ft - m) , we can calculate a_o and F, obtaining
a = 0.10 (n - m)" 1 , F - 2.9,
20
I-
o>
o
o - O.OOlm Nad
D- O.Olm Nad
A - O.lm NaCI
-4 -3
21
1.0
cre
o
o
fine
o o medium fine
10
JO
.05
10" 10
iov
SOLUTION CONDUCTIVITY (fl-m)H
sizes
22
a s = 0.08, F = 3.4,
for fine and medium-fine grain sizes, respectively. Using R co instead of R_O
The values of F calculated here are in good agreement with observed values
porosity, and imply an Archies law exponent of 1.2. Further comparison of our
data to the Waxman-Smits 1 model shall be made in a later section.
Figure 12 shows a decrease in T and m, with an increase in a c , The
possibly linear for small clay content, but then gradually decreases. Accom-
23
100 r
30
cr 20
10
0 2 4 6 8 10 12
WT % MONTMORILLONITE
24
panying the increase and leveling off of m is a decrease in background resis-
tivity, R The behavior of both m and R with increasing clay content can be
tion, but the clay in them does not significantly contribute to the polariza-
tent for only small volume fractions. The effect of clay content on RQ shall
an increased distance between selective zones, which does not seem probable.
a s - QVB
where Qy is the CEC per unit pore volume, and B is an empirical parameter
B = .001XI - aexp ( -) ,
where
25
F = Faraday constant,
a, y = experimentally-determined constants.
o 2
Qv is expressed in units of meq/m and B in units of m /ohm-meq. The clay
which was used was calcium montmorillonite (CMS STX1) which was obtained from
the Source Clays Depository (Clay Minerals Society) and has a CEC of 84.4
From the study of the effects of varying NaCl concentration with fine and
The study of the effects of varying the clay content with medium-coarse
assumes that the B parameter is independent of clay content. From the model,
we have
F/R - 0
-- -
v
Using measured porosities, resistivities, and solution conductivities, and
values of Qv calculated using the clay CEC and amount, B can be calculated as
a function of clay content. The results are presented in figure 14, where
agree with the other data and have been ignored in joining the data points
26
20 calculated using
calculated using
16
-f
I l2
o
2468 10 12
WT % MONTMORILLONITE
27
with a line. Note that the estimates of B (8 to 11) obtained in the preceding
paragraph for 3 percent clay with fine and medium-fine beads are in good
particles are close enough for diffuse zones to overlap and for conduction
resistivity data in figure 13. For 12 fi-m NaCl solution and 40 percent poros-
content.
a 0.83 0.6
28
Measurements by Waxman and Thomas (1976) support Group 1 as being correct.
given above. These values of B are not applicable to our results, however,
I II
since values of A given above are for Na ions rather than Ca ions. Accord-
I I
ing to Lorenz (1969), Ca ions have mobilities on kaolinite which are 40
by this amount, the second set of B values are obtained, also given above.
This second pair of values of B neatly bracket the values of B in figure 14
for high clay content. The best agreement is with the Group 1 values and
ures 15 and 16. Figure 15 shows observed spectra for 6 percent by weight Mg,
Na-montmorillonite (CEC = 65 meq/100 gms). The estimated chargeability for
this sample was only 0.02 compared to 0.16 for a sample with the same weight
29
1000
4/20 6%m m crs .003m . !A/mT2
m =0.022
T s '2.4
C =0.47
0 = U5
20. .. I 100
Q
X o:
o
.. 10
-2 -1 0 1
LOG FREQUENCY
Figure 15. Observed amplitude and phase for 6 percent by weight Mg, Na-
30
H h 1000
3/30 5%k .01mNaCl .!A/mT2
20.:. .. 100
a
x
o
9..
8..
7..
6.. *
5.
-3-2 -101
LOG FREQUENCY
Figure 16. Observed amplitude and phase for 5 percent kaolinite by weight,
31
Measurements on Natural Samples
descriptions available. The observed data for the samples are presented in
the Appendix. Samples 985-14 and 985-1IB were obtained from Gary Olhoeft
(USGS, Denver) and were collected at the Tony M. Mine in Garfield County,
Utah, from the Salt Wash Member of the Morrison Formation (Peterson, oral
commun., 1980). The samples were obtained from below the water table, and
consisted of grey, medium- to fine-grained sandstone, with clay and organic
material. There was no pyrite visible with 14x hand lens. Sample 985-14 had
a broad phase peak of 6 mrad centered at approximately 1 hz. Sample 985-1IB
had a noisy and weak phase peak of 2 mrad centered at about the same frequen-
cy. Data for sample 985-14 show a high-frequency phase response which is
thought to be a spurious effect associated with the measurement system rather
than the sample. This high-frequency response, which was observed with most
low frequency. This was thought to be due to drift in resistivity with time,
since the higher frequency data showed a decrease of similar amount when
repeated after the spectra were completed.
32
Sample 1602-21 was a grey, dirty sandstone of unknown origin. It dis-
played the nicely defined phase peak of approximately 30 mrad centered at 1/32
hz.
The measurements on natural samples cannot be used to test the applica-
is because the samples are so few in number and because ancillary mineralog-
ical textural data were not available. The data on sample 1602-21 and some of
the other samples demonstrate that clay-bearing sandstones can produce low-
frequency phase peaks which are consistent in form with peaks obtained with
NONLINEAR RESPONSE
were made using a ZONGE receiver in conjunction with an HP9825 desk-top calcu-
lator. The observed voltage waveform at each frequency was digitized using
the ZONGE receiver, and then transferred to the HP9825 where the harmonic
33
report here values of percent harmonic distortion at 2 and 3 times the funda-
mental frequency,
V
distortion = = x 100 percent ,
o
where
V = voltage at f,
f = transmitted frequency.
The ZONGE receiver has a 12-bit A to D, so the minimum noise level for our
ples of disseminated pyrite. We shall very briefly present the observed data
without any further discussion. Observed linear and nonlinear data for a
single pyrtie interface are presented in figures 17 and 18. A single inter-
frequency and for which the 2f harmonics are generally larger than the 3f
peak, while the 2f data are essential in the noise level. These data are
consistent with data of Anderson (1981). Note in figure 19 that the observed
thought to be due to drift effects rather than any unusual sample response.
34
10.0 10000
PYRITE INTERFACE , !A/mT2
1. 00 . . 1 1000
CVJ
o:
or
O
0. 10 . 100
0.01 10
-3-2-1 0 1 2
LOG FREQUENCY
Figure 17, Observed amplitude and phase for a single pyrite interface, 0.1
A/m2 .
35
PYF&TE INTERFACE . !A/mT2
0.01
-1 0 1
LOG FREQUENCY
2
Figure 18. Observed nonlinear data for a single pyrite interface, 0.1 A/m .
36
1000 1000
100. . 100
+ + + + + + -t-
Q
<
cr
o
10. ., 10
l. i i i i i 1
-3 -2 -1 0 J, 2 3
LOG FREQUENCY
Figure 19. Observed amplitude and phase for 10 percent pyrite disseminated in
o O
glass beads, pyrite grain size of 5 x 10"" Jm, 0.001m NaCl, 1.0 A/m.
37
4/15 10%Py .5cm .001m lA/mt2
* 2F
0.01
-2-101
LOG FREQUENCY
glass beads, pyrite grain size of 5 x 10~3m, 0.001m NaCl, 1.0 A/m2 .
38
The observed nonlinear response for clays is simple to describe because
dix. Nothing was observed which could be correlated with membrane polariza-
tion for the artifical samples. Sample Olhoeft 6 had response similar to
measurements made with a cation-exchange resin with a sharply increasing 3f
content which tended to increase with decreasing frequency and was rarely
greater then 0.1 percent. This behavior at low frequency is probably due to
typical electrode effects for measurements made on two different glass bead
blanks with no clay and with two different pairs of Ag/AgCl potential elec-
trodes. The response typically shows 2f and 3f components increasing with
decreasing frequency, and the 2f component usually greater than the 3f compo-
39
10.0
3/21 blank !A/mT2
* 2F
3F
i. 00..
z:
o
M O
\-
cn
I I
a
0. 10
0. 01
-3 -101
LOG FREQUENCY
Figure 21. Observed nonlinear data for glass bead blank, illustrating elec-
trode effects.
40
3/20 blank .003mNaCl !A/mt2
0. 01
-2-101
LOG FREQUENCY
Figure 22. Observed nonlinear data for glass bead blank, illustrating pro-
nounced electrode effects.
41
3. Measurements made on artifical samples suggest that the Waxman-Smits model
may be invalid for small clay concentration.
4. Measurements made on natural samples are of the same form as those ob-
served with artificial samples, and can be fit with the generalized Cole-
Davidson model*
42
REFERENCES
Madden, T. R., and Marshall, D. J., 1959, Induced polarization; a study of its
causes and magnitudes in geologic materials: Atomic Energy Commission
Report RME-3169.
12-13.
Olhoeft, G. R., and Scott, J. H., 1980, Nonlinear complex resistivity log-
ging: Society of Professional Well-Log Analysts, Transactions 21st Annual
Logging Symposium, p. T1-T18.
43
Roy, K. K., and Elliott, H. M., 1980, Model studies on some aspects of resis-
Sill, W. R., 1964, Induced polarization in clay bearing sandstones and the
Vacquier, V., Holmes, C. R., Kitzinger, P., and Lavengne, M., 1957, Prospect-
22, p. 660-687.
Waxman, M. H., and Smits, L. J. M., 1972, Electrical conductivities in oil-
bearing shaly sands: Society of Petroleum Engineering Journal, Transac-
44
APPENDIX
Observed Spectra
45
Natural Samples
46
Artificial Samples
47
/3ctO"
49 1000
20... 100
I Q
S
I
o
10. .. .. 10
9...
8...
7. ..
6...
5. 1
__Q
-101
LOG FREQUENCY
50
0.01
-3
-1 0 2
LOG FREQUENCY
1000
20... .. 100
I Q
:z <
IE o:
O
10.1 .. 10
9.
8.
7.
6.1
vv
5. 1
2-1012
LOG FREQUENCY
C-J
7;
DISTORTION
CO
en
to
oo
ui
O i
CD
~n 00
T
m
D CO
c: TI
Tl i\> 0
m
n
ro
1000
20... .. 100
I a
S <
IE o:
o
6...
5. 1
-3 -2-1 0 1
LOG FREQUENCY
1 ^^ mtm- 1 1 1 1 I
-- 3F
\
\. 00. *
O
1__1
STORT]
^%
,
Q
0. 101
3. 01
2-101
LOG FREQUENCY
55
1000
20. .. .. 100
I Q
S
I
o
10. -- 10
9.
8.
7.
6.1
5. 1
2-10123
LOG FREQUENCY
DISTORTION
O)
tx)
o
0
d-
fc. + * ro
O) ro
__ - , , , , , ***r ita^
20. i .100
S
1 ^i e- 0
S
IE
O * S
> *
8.. -X-
**
7..
6.. * ^ * <*
5. , , 1 1 1 '. 1 1 1
-3 -2-101
LOG FREQUENCY
*y %r-
/O m
DISTORTION S
CJ
Ul
o
T
0 r
fD !
+ O) +
O) ro
TI
4/30 Olhoeft 4 . 1m !A/mt2
100
20.
Q
I <
o:
X
o
X-
10
10.
9.
8.
7.
6.
5.
-3 2 -1 0
LOG FREQUENCY
(SI
DISTORTION
, i i. i = i i
20.. ^
100
:E
I Q
X C
-
o S
10. + + + ^^,fc * + + + + _h + + $ J :
10
9. **
8. ^y
r"!"
6* \A
TV
\f
^V
^ 4fe
^v ^ . J^.
*
3fr JUL
*
vT jy jy
"^CSr ~i^fc~
7.
6.
5. _____i i i i 1
2-1 0 1 2-3
LOG FREQUENCY
10.0
4/36 Olhoeft 5 . 1m !A/mt2
* 2F
+ 3F
1.00..
"2L
O
i
h-
tr
o
H
tfl
H-l
Q
0. 10
0. 01
-3 2-101
LOG FREQUENCY
1000
4/37 Olhoeft 6 . 1m !A/mt2
20. .. ,. 100
I Q
S <
X o:
o
10... ,,10
9...
8...
7...
6...
5. 1
-3 -2-10123
LOG FREQUENCY
64-
* 2F
+ 3F
0. 01
-1 0 1
LOG FREQUENCY
1000
4/38 Olhoeft 7 .1m lA/mt2
20. .. .. 100
I a
^:
IE
O
10.
9.
8.
7.
6.1
5. 1
-3 -2-101
LOG FREQUENCY
^ y %-*
/*>
DISTORTION g
CD
O
0
fD
O) ro
Tl
3-
ro
67
I t I K 1000
3/1 1602-21 0m/l . !A/mt2
20. .. .. 100
Q
<
o:
o
10. + + + +
+ + .. 10
9.
8.
7.
6...
5. 1 K 4 h 1
-3 -2 -1 0 1
LOG FREQUENCY
J10. 0
1.00
0. 01
-3 -2 -1 0
LOG FREQUENCY
7
A
DISTORTION g
U)
I ro
I\J
GO
3
o i
ci
~n
ro
m
D s
c CO I\)
Tl Tl
m
n
CD
.0 70
* 2F
3F
1.00..
o
I I
I-
5 O
I-
co
I I
Q
0. 10
0.01
-3 -2 -1 .0 1
LOG FREQUENCY
si * y
So
I DISTORTION
CO
ro
00
3
o i *
3
Cl (D
CL
0
m s + * CO
D i\)
c TI
m
n
-<
ro
3
ro
CO
10. 0 72
* 2F
i
+ 3F
1.001
o
I I
! H-
o:
^ o
) c/)
Q
0. 10
0.01
-2-1 0.1 2
LOG FREQUENCY .
10. 0 73
* 2F *
* 3F
1.001
'Z
o
I I
H-
o:
j^ f-
o
i C/)
ii
a
0.10
3.01
-2-1012
LOG FREQUENCY
10. 0
3/30 5%k .01mNaCl . !A/mt2
* 2F
3F
1.001
z.
o
t I
f-
K O
, h-
f CO
I I
a
0.10
0. 01
-3 -2 -1 0 1
LOG FREQUENCY